Providers and Scoring

Candidates are found by four providers which are ranked based on their score:

• Lexical matching

• Existing mapping matching

• Existing mapping lexical matching

• SNOMED CT Reference Set matching

• SNOMED CT lexical matching

Lexical matching:

Executes lexical term filtering queries based on the logic described in the Automap correlation threshold section. The score is returned by Elasticsearch's scoring algorithm which might be different for the same candidate with a different automapping configuration.

Only this provider is affected by the threshold, others are ignoring it and always return the same candidates.

Existing mapping matching:

Analyzes existing Mapping Set mappings and returns matches where the search code is present either as source code or target code (reversed mapping). Finding multiple existing mappings for a single search code has a higher score than a single existing mapping and regular mappings have a higher score than reversed mappings.

Existing mapping lexical matching:

Analyzes existing Mapping Set mappings and returns matches where the search code's term lexically (exactly and/or partially) matches either the source code's term or target code's term regardless of the actual source or target code values. Similarly to the existing mapping matching by code algorithm above, regular matches (matching the source term) have a higher score than reversed mapping matches. In addition, Existing mapping matches get higher scores than Existing mapping lexical matches.

SNOMED CT Reference Set matching:

This provider finds candidates which are map targets of a SNOMED CT Map type Reference Set member and returns its referenced component. Scores are the same for all results because they can not be distinguished accordingly.

SNOMED CT lexical matching:

The results are provided and scored based on the following logic:

Score	Description
10	exact case insensitive match between source term and an active SNOMED concept's preferred term
9.9	exact case insensitive match between source term and an inactive SNOMED concept's preferred term that has one of the following historical associations: SAME AS Association Reference Set (foundation metadata concept) REPLACED BY Association Reference Set (foundation metadata concept) 100% exact case insensitive match between source term and an active SNOMED concept's preferred term
9.8	exact case insensitive match between source term and an active SNOMED concept's synonym (where there is only one SNOMED concept with that synonym)
9.7	exact case insensitive match between source term and an inactive SNOMED concept's synonym (where there is only one SNOMED concept with that synonym) that has one of the following historical associations: SAME AS Association Reference Set (foundation metadata concept) REPLACED BY Association Reference Set (foundation metadata concept)100% exact case insensitive match between source term and an active SNOMED concept's preferred term
9.6	exact case insensitive match between source term (after stripping out NOS, etc.) and an active SNOMED concept's synonym
9.5	exact case insensitive match between source term (after stripping out NOS, etc.) and an inactive SNOMED concept's synonym that has one of the following historical associations: SAME AS Association Reference Set (foundation metadata concept) REPLACED BY Association Reference Set (foundation metadata concept)100% exact case insensitive match between source term and an active SNOMED concept's preferred term
8.0	Then you have another set like above that using POSSIBLY EQUIVALENT TO Association Reference Set (foundation metadata concept)